Significance of Criteria of

A study of the published data indicated that the plastic in- dices obtained on coals prior to coal 52 were inconsistent with those obtained for the co...
0 downloads 0 Views 419KB Size
In a survey of the carbonization characteristics of American coals, the U. S. Bureau of Mines reported values for the proximate analysis and plastic properties of the coals used and the physical properties of cokes produced. Sbtistical studies of there data made previously (6)showed a significant correlation between proximate analysis and physical Properties. . . Similar studies .reported here show that the plasticity indices obtained with Davis and Gieseler plastometers are highly correlated with the proximate analyses of the coals, and that the physical properties of the cokes produced can be calculated equally well from either plasticity indices or proximate analysis.

Significance of Criteria of

MOSG the tests used t o determine caking characteristic of coals, several empirical methods of determining various so-called plastic indices have received considerable attention. The plasticity of coal is a transient phenomenon, and the numerical values of the indices are functions not only of the coal but also of the appapatus and its mode of operation. If the latter are held rigorously constant, there is a highly significant correlation between the proximate analysis of a coal and the plastic indices obtained in either the Davis (2, 3 ) or the Gieseler (2, 5 ) plastometer as shown by the following analysis of data published by the U. S. Bureau of h/Iines in its survey of the carbonizing characteristics of American coals (9). A study of the published data indicated that the plastic indices obtained on coals prior t o coal 52 were inconsistent with those obtained for the coals tested later; this is possibly a result of changes in procedure introduced about the time this coal was tested, as described in a paper by Brewer and Triff ( 2 ) . For this reason the present report is limited t o a consideration of data for thirty-seven of the coals and blends numbered by the Bureau of Mines from 52 to 66 for which all the desired measures of plastic properties were given. These coals ranged in fixed carbon content, FC, on a dry, ash-free basis, from 59.2 to 82.5% with an average of 67.4%, and in ash content, A , from 2.9 t o 11.4% with an average of 5.5%. Statistical methods (7, 8) were used to determine the correlation between proximate analysis and the plastic indices of the thirty-seven coals and blends, and the regression equations relating the indices to the analysis were obtained by the method of least squares. Five indices measured by each of the two methods were studied. For the Davis plastometer they were:

A

H. H. Lowry

and

C. 0.lunge, Jr. C O A L RESEARCH L A B O R A T O R Y ,

CARNEGIE INSTITUTE OF T E C H N O L O G Y , PITTSBURGH, P A .

In all other cases the values of R, the multiple correlation coefficients, are greater than 0.70 and as great as 0.92 for TR,,,. Perfect linear correlation gives a value of unity to this coefficient. Also the values of the probable error of the calculated indices are small compared to the total range. It is unfortunate that these probable errors cannot be compared directly with the probable errors of the reported values of the indices, which are not given by the authors of the original data. From unpublished data kindly furnished by C. C. Russell of the Koppers Company, the probable errors of the characteristic temperatures measured with the Gieseler plastometer were calculated to be very close to 2.0” C.; this value was derived from eighty-three duplicate determinations. If it is assumed that the same value of the probable error is applicable to the Bureau of Mines data obtained with both plastometen, the probable errors of the calculated values are not much greater than those of the experimental. The low probable error of 3.0” C. for the calculated temperature of maximum resistance in the Davis plastometer is especially to be noted since this plastic index is of most general importance in calculating various measures of coke strength from the plastic indices, as will be shown later.

T R O = temperature a t which resistance begins TR, = temperature a t which resistance ends TR,., = temperature a t which resistance is a maximum TR,,,i, = temperature a t which the resistance is a minimum R,, = maximum resistance For the Gieseler plastometer:

TF = fusion temperature or temperature where pointer mpvement first becomes 5.0 dial divisions per min. T E = temperature at end of plastic range Ts = solidification temperature or the temperature where ppinter movement again becomes 5.0 dial divisions per mn. TP,,, = temperature a t maximum fluidity Fm.= = maximum fluidity

HE ash content of these coals had no significant effect on the Ttemperature a t which resistance begins, the temperature a t maximum fluidity, or the maximum resistance in the Davis plastometer. Likewise, the ash content was significant only in determining the solidification temperature in the Gieseler plastometer; this temperature wa5 lowered 2.5” C. for each increase in percentage of ash, which may be attributed to a stiffening of the “plastic” mass by the inert inorganic material. Although not shown in Table I, the plastic range, T s T p , as determined in the Gieseler plsstometer, would be demeased the same amount since it is obvious that

The relation of these indices to one another may become evident from examination of the data plotted in Figure 1 for a blend of coals from the Upper Kittanning and Pittsburgh seams (4). The results of the calculations of correlation coefficients and of the regression equations are shown in Table I. Where a constant in a regression equation was indicated to be zero, it was determined by statistical methods that the two variables are not significantly related; a direct least-square solution of the regression equation would yield a value of the constant greater or less than zero, but the probable error (P.E.) of the calculated value would not be significantly reduced. No signscant relation between the proximate analyses of the coaIs and the maximum determined in the Gieseler plastometer, was found. fluidity, F,,,

-

Ta

- TF = 65.6 - 0.145 FC

- 2.45 A

Although it is of unknown significance, an increase in ash content has an opposite effect on the plastic range, TR, - T R ~as, determined in the Davis plastometer, where

T R~ TRO= 85.5

308

- 0.206 FC + 1.29A

April, 1944

*

*

*

INDUSTRIAL AND ENGINEERING CHEMISTRY

309

Brewer and Atkinson (1) concluded Tnble 1. Results of Calculntions of Correlation Coefficients of Thirty-Seven Coals from their data on eleven coals and blends (No. 35-41 in the United States Bureau of -Detd. by Gieseler PlsstometerP D e t d . by Davis PlastometerMines survey) that the coke strength, as U* TRO TR/ TRmso TRmin Ems= TF TI$ TS TFmsz measured by the average of the percentYaw 413 492 478 30 446 413 464 455 435 gmoz 473 520 508 111 482 463 505 490 479 age retained on a 1.5-inch screen in the 894 427 425 416 6 432 gmin 396 468 456 shatter test and of the percentage of a R 0.75 0.87 0.92 0.75 0.84 0.84 0.70 0.73 0.85 a 262.6 348.1 345.6 -172.5 320.4 245.5 279.8 311.1 281.8 1-inch screen in the tumbler test ( S T ) ,inb 2.23 2.03 1.89 3.00 1.87 2.48 2.73 2.34 2 28 0 1.294 0.961 0 0 0 0 -2.45 0 creased as the plastic range determined in 7.2 4.2 3.0 9.6 4.4 5.8 10.1 8.2 5.2 $.E, the Davis plastometer decreased. The * ucdcd = a f bFC f c A . cokes were made in an 18-inch cylindrical retort a t a final temperature of 900' C. The correlation coefficient for their data was calculated to be -0.68, a value so were not stated by the authors of the original paper, the accuracy high that it could occur by chance only once in a hundred times. of the relations between coke strength and proximate analysis Also, the equation shown for their data was determined to be: cannot be determined. Correlation coefficients and regression ST = 124.2 0.77(T~/ T R O ) equations were calculated for these measures of coke strength However, the exact reverse is found to hold for the fortY-four and also for the 1.0-inch shatter index, T;, with the various coals numbered from 52 to 66 and the cokes produced from them plastic indices determined with the Davis plastometer and, for for which the necessary data are given: The coke strength indirect comparison, the p r o h a t e analyses of the same creases with increase in the plastic range; the correlation coefthirty-seven coals. The results of these calculations are shown ficient is 0.38 which, owing to the larger number of coals conin Table 11. sidered, is so high that it also could occur by chance Only Once in It is evident that the mean size after shatter and the 1.5-inch a hundred times; and the regression equation is shatter index can be calculated with slightly lower probable errors (higher correlation coefficients) from the analyses than from the ST = 53.78 0.197(T~, TRD) plastic indices. The only significant analytical factor is the fixed The latter relation is probably the more general since it was decarbon, and the only significant plastic index is the temperature termined on a brger number of coals which ranged in ST from a t which the coal develops maximum resistance, T R , ~ , . As 35.35 to 80.15 and in plastic range from 47" t o 106" c.,a8 CODshown in Table I, this plastic index can be calculated from the pared with the narrower ranges of ST from 56.85 to 74.40 and of proximate analyses with a probable error of only 3.0%. If the plastic range from 63' to 87" C. for the coals considered by Brewer method of partial correlation is used, and the correlation between and Atkinson. In any case, this measure of coke strength cannot Is and TR,,,.. is calculated for constant FC, the correlation coefbe closely predicted by measurement of the Plastic range of the ficient drops from 0.58 to the nonsignificant value of 0.05. Since coal from which the coke is made. there is no relation remaining between IS and T R,., when FC is held constant, the same factors of I s which are measured by TRmal must also be measured by FC. On the other hand, both REVIOUS studies showed that several other criteria of coke the tumbler indices can be calculated with somewhat lower prob'strength can be calculated from the proximate analysis of the able errors (higher correlation coefficients)from the plastic indices parent coal (6). For emmple, considering only data on cokes than from the proximate analyses of the coals. In addition to prepared a t 900" C. in the l&inch retort, the probable error of the TR,.,, S ~ S OTRO,the temperature a t which resistance develops, is calculated value of the mean size after shatter, Ms,was found to significant in determining the value of I;, the 1.0-inch tumbler 0.09 inch; that of the 1.6-inch shatter index, Is, 3.6%; and index; and the lower the value of this temperature, T R ~the , that of the 0.25-inch tumbler index, IT,2.0%. Since the experigreater the resistance to size degradation to 1.0 inch in the mental emors in determining these measures of coke strength tumbler test. It appears, however, that resistance to degradation to 0.25 inch in the tumbler test increases with TRO,which is in accord with the view that the breakage in the tumbler test to the larger sizes is to be attributed to a different mechanism from that which controls the amount of finer product of the tumbler test

-

-

+

-

Table Y

II.

Results of Calculations of Correlation Coefficients of Thirty-Seven Cokes

IS IT IC f bTRmaz f c T a -I-dRmam 1.97 78.5 72.7 58.3 2.30 88.9 77.3 69.8 1.40 40.3 66.0 23.7 0.62 0.58 0.70 0.72 -2.501 -140.9 -53.16 -226.8 0.00936 0.459 0.172 0.810 0 0 0.117 -0.246 0 0 -0,149 0 0.090 4.9 1.4 4.8

MS

Yealcd = a Yso ymaz

urnin

R

a

b

5P.E.

Ycslcd = a'

R

0

TEMPERATURE, 'C.

Figure 1. Plot of Data Obtained in Davis nnd Gieseler Plastometen with a Blend of Upper Kittanning and Pitbburgh Seam Coals (4)

a'

b'

0)

P.E.

0.68 0.523 0.0215 0 0.085

+ b'FC + o'A 0.62 9.13 1.030 0 4.7

0.42 62.85 0.1W -0.449 1.8

0.60 -19.09 1.149 0 5.6

310

INDUSTRIAL AND ENGINEERING CHEMISTRY

(6). This is further supported by the facts that the 0.25-inch tumbler index is the only one of these measures of coke strength that is significantly correlated either with the mBximum resistance developed in the Davis plastometer or with the ash content of the parent coal. I n previous work (6) another measure of coke strength was correlated with proximate analysis-namely, Rs, a constant measuring the uniformity of size after shatter. The range in value of R, for the thirty-seven coals considered was too small to warrant an attempt to calculate it from the plastic indices. Since the maximum fluidity as determined in the Gieseler plastometer was not significantly correlated with the proximate analysis, it appeared possible that inclusion of another term in the regression equation might improve the correlation coefficient and reduce the probable errors of the calculated measures of coke strength. It was found, however, that there was no significant correlation between the maximum fluidity and any of the strength indices No attempt was made to relate the other plastic indices determined by the use of the Gieseler plastometer to the several measures of coke strength since their correlation with proximate analysis was so similar to that found for the indices as determined by the use of the Davis plastometer.

Vol. 36, No. 4

ROM the preceding discussion, it does not seem unwarranted Fto conclude that knowledge of the various plastic indices considered, as determined by use of either the Davis or Gieseler plastometer, does not enable a closer prediction of several measures of coke quality to be made than is provided by the simpler proximate analysis. LITERATURE CITED

(1)

Brewer, R. E., and Atkinson, R. G., IND. ENG.CHmi., h

. 4 ~ H. D , ,

8 , 443-9 (1938).

Brewer, R. E., and Triff, J. E., I b i d , 11, 242-7 (1939). Davis, J. D., Ibid., 3, 43-5 (1931). Fieldner, -4.C., J . Inst. Fuel, 16, 5-20 (1942). (5) Gieseler, X., Gliickauf, 70, 178-83 (1934). (8) Lowry, H. €I.,Landau, H. G . , and Naugle, L. L., Trans A m .

(2) (3) (4)

Inst. Minino M e t . Enors.. 149.297-330 (1942). (7) Mills, F. C., 'Statistical Methods", rev. ed., New York, Henry Holt. 19311. (8) Waugh, A. E., "Elements of Statistical Method", New York, McGraw-Hill Book Co., 1938. (9) R7ilson,J. E., and Davis, J. D., U. S. BUI.Mines, Tech. Papw 637 (1942). PRESENTED before the Division of Gas and Fuel Chemistry a t t h e 108t,h Meeting of the A M E R I C ACHDMICAL ~J SOCIDTY, Pittsburgh, Pa.

BICYCL0[2,2,1] EPTANE and

BICYCL0[2,2,l]-%HEPTENE Bicycl0[2,2,1]-2-heptene is formed by heating technical dicyclopentadiene with ethylene in a bomb at 200' c. and a t pressures up to 120 atmospheres. Presumably the dicyclopentadiene depolymerizes to cyclopentadiene, and this reacts with the ethylene under the reaction conditions to give bicycloheptene. Bicyclo[2,2,l]heptane is formed by hydrogenating the bicycloheptene at 50' C. in the

presence of a nickel catalyst. A small amount of methylicyclohexane is formed at the same time. Both of the bicyclo compounds are white crystalline solids. A. S.T.M. octane numbers have been determined on blends of 2,2,4trimethylpentane, n-heptane, and the hydrocarbon. The bicycloheptene has a blending octane number of 95 * 5; the bicycloheptane, a blending octane number of 56 * 5.

CHARLES L. THOMAS Universal Oil Products Company, Riverside, Ill.

T

HE formation of bicyclo [2,2,1]-2-heptene by the reaction of cyclopentadiene with ethylene was recently reported (7). The reaction was carried out at 190-200' C. and at pressures to 395 atmospheres. A considerable amount of the bicycloheptene was desired for study, but the rather high pressure of 395 atmospheres was a deterrent to making large preparations, The preparation and storage of large quantities of cyclopentadiene also presented difficulties. In the present work it was found that both of these difficulties could be overcome by charging technical dicyclopentadiene and ethylene to the pressure vessel. Apparently the dicyclopentadiene cracks to cyclopentadiene, which reacts with the ethylene. Under the conditions used, the maximum pressure encountered was 120 atmospheres. The bicycloheptene was readily hydrogenated catalytically with a nickel catalyst. The expected bicyclo [2,2,l]heptane was obtained, together with a minor amount of methylcyclohexane. Both of the bicyclo compounds had been previously prepared by other methods (8). The combustion characteristics of these materials in internal combustion engines is of interest, especially when compared with

the combustion characteristics of hydrocarbons that might, be considered related to bicycloheptane. The octane numbers of such hydrocarbons follow : Compound

Octane No.

2 3-Dimethylpentane

90 (8) 82 (6 90 ( 2 ) :80 (61

2:4-Dimethylpentane n-Hexane 28 ( 0 ) Cyclohexane 80 (8) 0 Blending octane number.

Compound n-Pentane Cyclo entane BicyoEheptane

Octane No.

2*86"

66

Bicycloheptane contains two tertiary carbon atoms and might be expected to resemble 2,3- or 2,Pdimethylpentane and have an octane number in the 80-90 range. Or since it can be considered as containing two cyclopentane rings, it might be expected to resemble cyclopentane and have an octane number of about 85. Or since n-pentane has an octane number of 64 and this is raised, on conversion to cyclopentane, to 85, then further cyclicizing might be expected to give a still higher octane number. Bicycloheptane can also be related to cyclohexane which has an octane number of 80. It might be expected to have a somewhat Kigher octane number if the n-hexane series is valid, since nhexane with 28 goes to 80 on cyclicixing t o cyclohexane. A con-